computing-related grant programs nih and...
TRANSCRIPT
1
Computing-Related Grant
Programs
NIH and NLM
Charles P. FriedmanSenior Scholar, NLMCRA Academic Careers WorkshopFebruary, 2006
2
Overview
� Introduction to NIH and NLM� NIH = “National Institutes of Health”� NLM = “National Library of Medicine”
� The computational and “informatics” scene at NIH� A primer on NIH funding� NLM grant programs
� See also “extra slides” we won’t cover this AM
3
National Institutes of Health
� The steward of medical and behavioral research for the nation
� One of the world's foremost medical research centers
� Intramural and Extramural research� Research Grants and Contracts; Training
programs
� $29 billion Congressional appropriation in FY 2005
4
National Institutes of Health
� 27 Institutes and Centers (“ICs”)� “Categorical” Institutes - Cancer, Heart, Lung &
Blood, Kidney, etc.� “Non-Categorical” Institutes - NIGMS, NHGRI,
NCRR, NIBIB, NLM
� 77,900 grant applications received in FY 2005
5
NIH and Computer/Information Science
� Historically, NLM was the “IC for computing”� Then three epochal events changed things:
� 1999 Biomedical Information Science and Technology Initiative (BISTI) Report: http://www.nih.gov/about/director/060399.htm
� 2002 NIH Roadmap:http://nihroadmap.nih.gov� 2003 Cancer Biomedical Informatics Grid (caBIG):
https://cabig.nci.nih.gov
� Now computing and informatics at NIH is spreading faster than it can be tracked
6
So What Is Informatics?
Basic Informational & Computational Sciences
An Application Domain
“<Domain> Informatics”
Information ScienceComputer ScienceStatisticsCognitive ScienceOrganizational Scienceetc.
Health CareBiologyPublic HealthClinical Researchetc.
7
An “Informatician” is Different…
Basic Informational & Computational Sciences
An Application Domain: Biology
or Medicine
What an Informatician Knows
What an Information Scientist Knows
What a Biologist or Physician Knows
8
The Question of Training…
The trade-offs between:� “Domain generic” computer or information
science training� Informatics cross-training with domain
immersion
� Some CS/IS graduates may want postdoctoral informatics training
9
NIH Roadmap
Pan-NIH initiativesComputing threaded through all of them� New Pathways to Discovery: Biological
pathways, molecular libraries, bioinformatics, nanomedicine, computational biology
� Research Teams of the Future: Interdisciplinary; team science; high risk research
� Reengineering Clinical Research: Research networks, outcomes assessment, translational research
10
National Centers for Biomedical
Computing� Grew out of BISTI and Roadmap� Networked national effort to build a
computational infrastructure for biomedical research� From basic research in computational science to
providing computational tools and resources for biomedical/behavioral research
� $4,000,000 per year for 5 years� 4 funded in 2004; 3 more in 2005
11
NIH Sources of Information
Computing at NIH is no longer localizable…� NIH Institutes & Centers http://www.nih.gov� NIH Roadmap http://nihroadmap.nih.gov� BISTI http://www.bisti.nih.gov� NIH Guide weekly listserv
http://grants1.nih.gov/grants/guide/listserv.htm
� CRISP database of funded projects http://crisp.cit.nih.gov
12
The Structure of NIH Funding
Opportunities
� Program Announcements (PA): Areas of persistent interest; recurring submission dates
� Requests for Applications (RFA): One-shot but may be reissued
� PAs and RFAs may be issued by a single IC or multiple ICs
� There are multiple grant mechanisms; e.g.:� Rxx = research grants� Kxx = career development awards
13
Some Rudiments of NIH (and NLM)
Grant Procedures
� After submission, grant applications are assigned to ICs for oversight and payment, should the application be funded
� Assignment is distinct from merit review!� Two modes of merit review assignment:
� Review by the study sections under the assigned ICs (~ single IC oppys)
� Review by study sections administered by the Center for Scientific Review (~ multi-IC oppys)
14
Grant Review Outcomes
� PRIORITY SCORES on 100 (best) to 500 (worst) scale� 100-150 (most likely to be funded)� 150-170 (seldom funded)� 170-200 (rarely funded)� 250-500 (never funded; sometimes don’t receive scores)
� Priority scores are the primary basis of award decisions
� Most applications can be revised and resubmitted twice
15
Some Hints and Tips
� Generate a idea� Find potentially compatible receptor sites (PAs and
RFAs)� Hard to “create” a receptor site at NIH� Program scopes are broad but inflexible at boundaries
� Write a “one pager”� Contact the program officer (you may be referred to
a better receptor!)� Visit if possible� Don’t expect funding the first time
16
Funding Activities at NLM
� Biomedical Informatics is NLM’s research domain� We don’t fund generic CS/IS� Projects must be domain-immersed
� Grants (Mostly Recurring PAs)� Contracts (Mostly Episodic)� Training (Recurring)
� Fellowships for informatics researchers, informationists� 18 University-based Training Programs� Lister Hill Center intramural training opportunities� NCBI short term training� Woods Hole Informatics Short Course
� Career Development Awards (Recurring)
17
NLM & NIH - snapshots
NIH in FY 2005� $29 billion from
Congress, $19.8 billion in research grants
� 77,904 grant applications received, 5.6% increase over FY 2004
� 27,700 active grants� Average research grant
= $355K, 3.7 years
NLM in FY 2005� $317.9 million for NLM,
$54.2 million for EP grants
� 500 grants assigned to NLM, 80% reviewed by NLM
� 150 active grants� Average research grant
= $390K, 4.5 yrs
18
NLM Informatics Research Areas of Interest
(PA06-094): Must be Domain-Immersed!
� Information and knowledge processing, including natural language� Processing, information extraction, integration of data from
heterogeneous sources or domains� Tools for analyzing and/or storing very large datasets� Knowledge representation, including vocabularies, ontologies,
simulations and virtual reality� Linkage of clinical and genomic information� Innovative uses of information technology in health care delivery or
public health� Efficient management and utilization of information and data� Knowledge acquisition and management, process modeling, data
mining, acquisition and dissemination� Human-machine interaction, including interface design, use and
understanding of health related-information� Uses of information technology to enhance learning, retention and
understanding of health-related information
19
Career Development Awards
� NLM is focusing on new “Pathways to Independence” Awards (K99/R00: PA-06-133)
� One award provides up to five years of support in two phases:� 1-2 years of mentored support for postdocs� Up to 3 years of independent support contingent on
acquiring a tenure-track position� 2nd phase does not require a separate competing
application
� Minimum 75% supported research effort in both phases
20
NLM’s 18 Training Siteshttp://www.nlm.nih.gov/ep/GrantTrainInstitute.html
21
Research Grant Programs
� R01 – Traditional ‘investigator-initiated’ research projects – hypothesis driven research; average $350,000/yr for 3 yrs
� R21 – Exploratory/Developmental research grants – feasibility & proof of concept; “high risk” $275,000 over 2 years
� R03 – Small research projects – no prelim data needed; $50,000/yr for 1 or 2 yrs
22
Research Grant example
� R01 (Vanderbilt): Clinical and MicroarrayData Predict Lung Cancer Outcomes. Compare rule-based induction to support vector and artificial neural networks; develop and validate backward chaining rule induction as a means of ‘semi-supervised’ learning
23
Research Grant example
� R01 (Columbia): Discovering & Applying Knowledge in Clinical Databases. Develop and test methods to mine a clinical data repository, exploiting latent associations and methods of data representation to improve classification of cases characterized by complex, missing or inaccurate data, using nearest neighbor and case-based reasoning
24
Research Grant example
� R01 (Michigan): Computational Approaches, Protein Sequence Analysis. Develop automated knowledge extraction techniques to find molecular biology data expressed as assertions in published articles through multi-stage parsing and preposition templates, making inferences on molecular biological context
25
Exploratory Research Grant
� R21 (Indiana). Syndromic Surveillance Data Exchange & Analysis. Incorporate geospatial clustering with temporal information to develop pattern recognition algorithms for de-identified patient data at hospitals in 2 cities
26
Small Research Grant
� R03 (UCLA) Content-based Neuro Image Classification: Perform automated classification of tumor images based on principal componential analysis and provide ability to query by sub-regions of an image
27
Points of contact at NLM Extramural
Programs
� Dr. Milton Corn – NIH Roadmap initiatives � Dr. Valerie Florance – Research in clinical
informatics and information sciences; IAIMS; career development awards
� Dr. Charles Friedman - Research in bioinformatics; Training Programs, Fellowships, National Biomedical Computing Centers
� Dr. Hua-Chuan Sim – KM/AI grants; R03 Small grants; Conference grants; Scholarly Works grants, SBIR/STTR
28
Basic NLM Contact Information
� [email protected]� 301-594-4882 for Florance, Friedman� 301-496-4253 for Sim� 301-496-4621 for Corn� http://www.nlm.nih.gov/ep for program
announcements, FAQ, links to application forms, list of funded projects
� http://www.nlm.nih.gov/grants.html for intramural fellowships, research contract information
29
Additional Slides Follow
30
Grants for Small Businesses
� SBIR/STTR phase 1 and 2 grants for companies that want to bring a product to market
� For NLM, priorities are the same as for research grants in biomedical informatics
� Fast-track possible – combine phase 1 and 2 in a single application
31
NIH Review Criteria for Research Grants
� Does it address an important problem?Significance
� How sound is the scientific environment?
Environment
� Does the PI/team have appropriate training and experience?
Investigator
� Are the aims, concepts, methods and/or outcomes novel?
Innovation
� Is the conceptual framework sound?� Is related work discussed?� Are potential problems recognized?
Approach
Questions in Reviewer’s mindCriterion
32
NLM Grant Deadlines
1-December1-August1-AprilSBIR/STTR
5-December5-August5-AprilFellowships
1-November1-July1-MarchRevised Applications
1-October1-June1-FebruaryResearch,
Career, Resource
Grants
Cycle 3Cycle 2Cycle 1
33
Review Steps for Grants
� REVIEW STEP� Received at CSR
�
� 1st Review by BLIRC �
� 2nd Review by BOR �
� Final Decision, NLM �
� TIMETABLE� Feb 1 (Jun 1, Oct 1)� May (Nov, Mar)
� Priority scores @ 1 week� Summary statement @ 6 week
� Sept (Feb, May)
� @ Nov (Apr, Aug)� Notice of grant award mailed
34
Components of Good Proposals
� Responsive to the program announcement� Clearly-stated goals and methodology� A detailed work plan that fits the stated
goals, with timeline and milestones� References to the published literature
and/or state of the art tools & systems
35
Components of Good Proposals
� Research grant proposals require a testable hypothesis and/or an accepted, rigorous research methodology
� Exploratory/developmental grant proposals need milestones and expected benefits
� Involvement of experts from the application domain (biomedical scientists, clinicians)
36
Components of Good Proposals
� Interesting pilot data� Evidence that problems & contingencies have
been considered� Key personnel with relevant expertise and
adequate FTE dedicated to the work� Letters of agreement from named
collaborators & consultants